452994 - Reloading a bug when a bug has been added to the "Depends on" list triggers edit mode

Status: RESOLVED WONTFIX

(Bugzilla :: User Interface, defect, P3)

Description

[Smokey Ardisson (offline for a while; not following bugs - do not email)]

I've seen this happen a couple of times now, but I'm not sure what's really causing it (i.e., what's happening between step 1 and step 2 below).

STR:

1) Open a bug that's likely to have bugs added to its "Depends on" list.
2) Some time later, return to the the tab containing that bug
3) Reload

AR: "Depends on" list is in edit mode (the text box for entering bugs has re-appeared and the "edit" link has vanished)

ER: "Depends on" list remains in "read" mode (line of text links, plus an edit link at the end).

I *think* what triggers this is that in the interim (between step 1 and step 2), a bug has been added to the "Depends on" list, but I haven't been paying enough attention when I see this to be sure.

Comment 1

[Smokey Ardisson (offline for a while; not following bugs - do not email)]

Yeah, that's what seems to be happening.  When reloading bug 452731 after filing this bug, bug 452731 reloaded with the "Depends on" list in edit mode.

Comment 2

[Max Kanat-Alexander]

Yes, that is likely what has happened, since our only way to know whether or not you should be in edit mode by default (such as when you click Back) is whether or not the field is different from the bug's stored value.

Comment 3

[Guy Pyrzak]

This is actually expected behavior and should be resolved as a "won't fix", I'll explain why, but first let me explain the steps a bit more clearly.

1. open a bug A which has some other bug Z depending on it
2. in a new tab open another bug B
3. make B block on A
4. Go back to the tab with bug A
5. Refresh bug A

"bug" occurs.

Here is what is actually happening:
Your browser understands "bug Z depends on bug A" when you initially loaded the bug but when you refresh, the html that comes back actually says "bug Z and bug B depend on bug A", however, the browser remembers what you had in the field from before the refresh occurred, the browser (trying to be smart) understands that you had filled it in yourself, and therefore refills in the field with "bug Z depends on bug A". 

If this occurs the browser is then the Bugzilla code is supposed to inform you that if you save bug A with the current state, it will CHANGE bug A and remove the depends on bug B. It does this by changing the field into a text area as if you had clicked edit (which might not be salient enough). This is also used to help if people go to a bug, enter some invalid data, get an error and click back. If this mechanism wasn't there the user would not see the changes they had made, even though they are there.

There is even a bug on BMO about "(edit)" not working after an error for the old version of BMO, but I don't know where it is right now.

Comment 4

[Dave Miller [:justdave]]

Yeah, seeing the full description, I agree, this is intended behavior.  Personally, I've gotten into the habit of shift-reloading a bug if I've left it open while editing other things that touch it, mainly because the radio buttons used to get screwed up if you did a normal reload (that particular problem is gone now, but I guess that's a good habit to keep).

Comment 5

[Serge Gautherie (:sgautherie)]

I "disagree" about the WontFix:
this is a regression, simple reloading did work (for me) before this update;
now, a forced reload does loose the real editing that was done (on other fields).
At the very least, it triggers "edit mode" on a field on which I have done no editing: confusing.

(NB: Fwiw, I don't recall having issues with radio button.)

*****

You can also see the opposite bug on the CC list Add field for example:
Edit, type something, simple Reload,
the CC list goes back to "view mode" whilst what you typed is still in the Add field ... confusing, yes ! :-/

My feeling is not that the "Depends on" field (only) has an issue,
it's that the whole new ui behavior wrt view/edit is "broken"/regressed in case of Reload...

Comment 6

[Dave Miller [:justdave]]

If you're calling this a bug, it's a bug in your browser (and several of them, since they almost all behave the same) in not properly keeping track of what you actually edited when deciding what in the form to cache and what to refresh from the newly-retrieved HTML.

And I can guarantee you that Firefox did the same thing on the old version of Bugzilla, it's just that Bugzilla didn't go pointing it out to you before like it does now (you would have had the same thing in the depends box that you did before refreshing, and submitting the bug would put it back that way, even though it had been changed in between).

Comment 7

[Dave Miller [:justdave]]

And the CC thing is bug 453932